In a project, I have documents identified by UUIDs. For the lack of a better strategy, I decided to partition them based on the value of the UUID identifier. For example, my first partition could be 00000000-0000-0000-0000-000000000000 - 0fffff-ffff-ffff-ffff-ffffffffffff, the second one 10000000-0000-0000-0000-000000000000 and 1fffff-ffff-ffff-ffff-ffffffffffff, and so on. [Read More]

For a project, I need to store large amounts of documents and retrieve chunks based on their relevance. My current solution is a small embedded Chroma, and I wanted something more scalable. Enters: Elasticsearch. [Read More]

Just published on my Substack: Using AI to diagnose medical problems - We are not there yet but with the right approach… [Read More]

This document explores a multi-layer approach to ingesting and retrieving a large amount of documents in RAG applications. [Read More]

I just posted to my Substack a quick guide to run Stable Diffusion locally. It covers the basics, from setup to inpainting to correct errors. [Read More]